Differences

This shows you the differences between two versions of the page.

--- resources:grogu_setup [2024/03/06 10:10] – [Installed MATLAB] admin
+++ resources:grogu_setup [2024/03/07 10:32] (current) – [Third Try] admin
@@ Line 10: / Line 10: @@
 ====== Third Try ======
+-02-20
 In summary, Second Try to install VPixx 3DPixx was doomed by having copied the wrong file to ''/usr/share/X11/xorg.conf.d/xorg.conf''.
@@ Line 366: / Line 368: @@
 VPixx (Lindsey Fraser from technical support responding on behalf of Sophie Kenny) replied to me within an hour!  Based on the detailed email instructions, I did both of these things:
-. Changed line 62 of DatapixxImagingStereoDemo.m
+===== 1. Changed line 62 of DatapixxImagingStereoDemo.m =====
 Changed the argument from "2" to "5":
@@ Line 383: / Line 386: @@
 This made it so that the LCD glasses at least started flickering.  They were badly out of phase or frequency, so that there was no stereopsis at all, however.
-. Deleted ''/usr/share/VPixx Software Tools/'' directory and installed latest version of VPixx software tools from their website ([[https://docs.vpixx.com/download/|https://docs.vpixx.com/download/]]).
+===== 2. Deleted /usr/share/VPixx Software Tools/ directory and installed latest version of VPixx software tools from their website =====
+([[https://docs.vpixx.com/download/|https://docs.vpixx.com/download/]]).
 This installed everything to ''/usr/local/share/VPixx Software Tools''.
@@ Line 508: / Line 514: @@
 MATLAB was installed to ''/usr/local/MATLAB/R2023a''
+Had to do ''restoredefaultpath;matlabrc'' first time opening MATLAB.  Then I clicked "Save" in the path dialog.  Hopefully that will prevent future can't find path errors.
 ===== Installed PTB for MATLAB =====
+I think when I installed the MATLAB-specific version from NeuroDebian, it overwrote the Octave-specific version.
+I did this:
+<code bash>
+sudo apt-get install matlab-psychtoolbox-3
+</code>
+Instead of making symlink to Datapixx.mexa64, just put the directory holding the correct file at the top of my MATLAB path (''/usr/local/share/VPixx Software Tools/Software Tools/Matlab/mexfiles/linux/matlab/Datapixx.mexa64'').  Before this, a copy of ''Datapixx.mexa64'' from the psychtoolbox-3 subdirectories was being used by MATLAB, and it wouldn't do ''Datapixx('Open')''.  Now it works.
+''ProceduralGarboriumDemo.m'' and ''VBLSyncTest.m'' work great.
+===== Tested 3DPixx with MATLAB PTB =====
+Tried ''HolesCrowd_PTB_3StimCrowd_TEST.m'' with the VPixx LCD glasses.  It was somewhat better than with Octave.  There was still a dark band at the bottom of the screen, but it stayed at the bottom, so that its problem is that it causes a distraction, but not interference with the stimuli themselves.
+The glasses did flicker noticeably, which is a bigger problem.
+I checked the script, and this is the script that has ''Datapixx('SetVideoStereoWaveform',0)'' like I used with Octave last Friday.
+**Update** (2024-02-06 13:20): The display had become set to 85 Hz somehow.  I reset it to 119.8 Hz, and the flickering went away.  The dark horizontal bar persists, however.
+==== The Datapixx code lines from the script HolesCrowd_PTB_3StimCrowd_TEST.m ====
+I am 100% sure that I copied these lines from DatapixxImagingStereoDemo.m in 2014:
+<code matlab>
+% Hardware stereo buffers are a great idea, but they just seem to be broken on so many systems.
+% Set to 1 to try it out, or set to 0 to implement software frame-alternate buffers.
+useHardwareStereo = 0;
+[... many lines later: ...]
+% Prepare pipeline for configuration. This marks the start of a list of
+    % requirements/tasks to be met/executed in the pipeline:
+    PsychImaging('PrepareConfiguration');
+    % Tell PTB we want to display on a DataPixx device:
+    PsychImaging('AddTask', 'General', 'UseDataPixx');
+    % Decrease GPU workload
+    PsychImaging('AddTask', 'AllViews', 'RestrictProcessing', CenterRect([0 0 512 512], Screen('Rect', screenNumber)));
+    % Enable DATAPixx blueline support, and VIEWPixx scanning backlight for optimal 3D
+    Datapixx('Open');
+    %Datapixx('EnableVideoScanningBacklight');       % Only required if a VIEWPixx.
+    Datapixx('EnableVideoStereoBlueline');
+%    Datapixx('SetVideoStereoVesaWaveform', 2);      % If driving NVIDIA glasses %commented out 2024-02-28 by adc
+     Datapixx('SetVideoStereoVesaWaveform', 0);    % If driving 3rd party emitter %uncommented 2024-02-28 by adc
+%   Datapixx('SetVideoStereoVesaPhase',0); % Added 2024-02-28 by adc to try getting rid of drifting horizontal band
+    Datapixx('RegWr');
+    % Consolidate the list of requirements (error checking etc.), open a
+    % suitable onscreen window and configure the imaging pipeline for that
+    % window according to our specs. The syntax is the same as for
+    % Screen('OpenWindow'):
+    if useHardwareStereo == 1
+        [w, wRect]=PsychImaging('OpenWindow', screenNumber, 0, [], [], [], 1);
+    else
+        [w, wRect]=PsychImaging('OpenWindow', screenNumber, 0);
+    end
+    % There seems to be a blueline generation bug on some OpenGL systems.
+    % SetStereoBlueLineSyncParameters(w, wRect(4)) corrects the
+    % bug on some systems, but breaks on other systems.
+    % We'll just disable automatic blueline, and manually draw our own bluelines!
+    if useHardwareStereo == 1
+        SetStereoBlueLineSyncParameters(w, wRect(4)+10);
+    end
+    blueRectLeftOn   = [0,                 wRect(4)-1, wRect(3)/4,   wRect(4)];
+    blueRectLeftOff  = [wRect(3)/4,   wRect(4)-1, wRect(3),     wRect(4)];
+    blueRectRightOn  = [0,                 wRect(4)-1, wRect(3)*3/4, wRect(4)];
+    blueRectRightOff = [wRect(3)*3/4, wRect(4)-1, wRect(3),     wRect(4)];
+    % Initially fill left- and right-eye image buffer with black background
+    % color:
+    if useHardwareStereo == 1
+        Screen('SelectStereoDrawBuffer', w, 0);
+        Screen('FillRect', w, BlackIndex(screenNumber));
+        Screen('SelectStereoDrawBuffer', w, 1);
+        Screen('FillRect', w, BlackIndex(screenNumber));
+        Screen('Flip', w);
+    else
+        Screen('FillRect', w, BlackIndex(screenNumber));
+        Screen('Flip', w);
+    end
+</code>
+==== Tinkering with HolesCrowd_PTB_3StimCrowd_TEST.m ====
+Commented out the line that skips the PTB sync tests.  I had included this line because it was the only way I could get the stereopsis working in 2014, I think.  But no reason to keep it now.
+The line (line 6):
+<code matlab>
+Screen('Preference', 'SkipSyncTests', 1);
+</code>
+Ironically, perhaps, there are now some extra flashes and flickers that seem to happen when the background of the display changes, although fortunately not when the letter stimuli appear.
+I tried uncommenting-out (turning on) this line:
+<code matlab>
+   Datapixx('SetVideoStereoVesaPhase',0); % Added 2024-02-28 by adc to try getting rid of drifting horizontal band
+</code>
+Result: band moves around now = worse than before.  Will comment the line out again.